An approach to estimating the effects of nuclear attack at the local level is outlined; the methodology developed to estimate these effects is described; and the results of applying this approach in a pilot study for Houston, Texas are presented. Population protection systems in which shelter location is fixed and systems in which shelter location and type are optional are evaluated for single-weapon attacks ranging from 0.1- to 100-Mt against both the at-home and at-work population. One fixed shelter system is considered: a system based on data from the National Fallout Shelter Survey. The location- optional systems considered use in-city shelters with strength and location determined by a damage-limiting shelter-allocation model. The results of the analyses of population-protection systems suggest that: (1) NFSS shelter, although effective against fallout from a distant attack, may be counter- productive against direct attack; (2) a system of dispersed shelters promises huge life savings at low cost and warrants further study; and (3) the in-city damage limiting systems while expensive can be efficient against a single-weapon 10-Mt attack.